Ship an AI SaaS
in a day.
The open-source starter in plain Python and HTML. No React, no build step, no node_modules.
The entire kit is 559 lines. Read every file before lunch.
Free core, MIT license · Pro launches October 13
pip install -r requirements.txt cp .env.example .env uvicorn app.main:app
Then open http://localhost:8000, sign up, and chat.
What's inside
- —
Auth
Email and password, PBKDF2-hashed, signed HTTP-only session cookies. Standard library only.
- —
AI chat
OpenAI-compatible client. One env variable points it at OpenAI, Fikra, or any compatible provider.
- —
PWA
Installable, offline shell. A manifest and a service worker, nothing more.
- —
SQLite
Single file, created on first run. No ORM, no migrations framework.
- —
Frontend
One HTML page, Tailwind via CDN, vanilla JS. No build step.
What we left out
- No React, no TypeScript, no node_modules
- No ORM, no Docker Compose, no Redis, no Celery, no CI YAML
- No linter config forest
Every file in the repo exists because your SaaS needs it. That's the whole point.
Free AI credits included
Anza ships pointed at Fikra, an OpenAI-compatible inference API. Your kit includes $1 in free credits, and credits let you choose whichever model you want instead of locking you into one model's free tier.
When your app outgrows experiments, the same account scales without changing a line of code: self-serve from $1, Growth at $50/mo, Scale at $138/mo.
Free vs Pro
| Free (MIT) | Pro — $59 one-time | |
|---|---|---|
| Auth, chat, PWA, SQLite | Yes | Yes |
| Payments wiring (Paystack + Stripe) | No | Yes |
| Email sequences and admin dashboard | No | Yes |
| Lifetime updates | No | Yes |
Pro launches October 13.
Leave your email and get the launch price plus the buyer's guide on day one.